Art and Science of Data Visualization (Part 2)

Just a little more theory and background on Data Visualization, to eat up the rest of your weekend. Unlike the McCandless video, this video will take some time investment – but well worth it, if you want to charge your batteries on this topic.

(via Flowing Data) From earlier this year, Martin Wattenberg gives a talk on data and visualization. Some interesting points:

Data visualizations work well with large data sets
Interesting insight comes from borrowing ideas from different knowledge areas
Visualizations and analysis can help text “data sets” just as much as numerical data sets

Plus, a really interesting injection of artistic sense and sensibility into the “science” of data analysis – breaking WIRED magazine covers into component colors wheels, and seeing editorial patterns following patterns. Wattenberg’s summation on this particular visualization technique:

… an interesting phenomenon – the idea that you can visualize color.

I know what he means, it just sounds like he’s struggling for words to express a concept that is simple to understand visually; sort of a mini-proof of the “picture tells a thousand words” idea that is the fundamental reason why well-done data visualizations are amazingly effective.

Presentation fans, take note: Wattenberg also provides multiple examples of effectively reinforcing a discussion point by doing really interesting visualizations with openly available tools on the internet. Makes me question my “firmly held belief” that presenting live software to a large group is a a recipe for disaster. This is one of my personal takeaways from the video – live interaction helps the audience better understand the insights that they can get when working with the data. A least it’s better than a room full of bobbleheads when the “drilldown” buzzword pops up.

By the way – the MITWorld site has a huge collection of deep-dive videos; more on data visualization and empirical analysis, of course, but on a wider range of topics as well.